Introduction

The conundrum of non-adaptive ancient gene loss

Origination and extinction events are integral to the evolutionary diversification of life at the organismal and molecular level. Much progress has been made in understanding how new genes originate and how these events translate into organismal evolution (Ohno 1970; Long et al. 2003; Zhou et al. 2008; Hahn 2009; Innan and Kondrashov 2010; Schlötterer 2015; McLysaght and Hurst 2016; Rodrigo and Fares 2018). These studies have also shed light on the frequency and dynamics of gene extinction at the microevolutionary time scale, revealing a continuous background decay and loss of most nascent young genes because of their functional redundancy or deleterious dosage effects (Lynch and Conery 2000). Extinctions, however, have now also been well documented for genes with millions of years of evolutionary history. In recent years, comparative genomics has begun to deliver ultimate insights into the abundance of ancient gene loss (Albalat and Cañestro 2016). As an example, the last common ancestor of eumetazoan animals was inferred to have been in the possession of over 7500 gene families. Most of these are still conserved in vertebrates but over 15% (~ 1300) have been lost in early protostomes (Albalat and Cañestro 2016).

Several trajectories have been found to account for the loss of ancient genes. For one, gene loss can be adaptive (Albalat and Cañestro 2016), such as the loss of genes coding for receptor proteins that are utilized by pathogens during cellular infection (Hedrick 2011; Albalat and Cañestro 2016). Another category of gene loss arises through the regression of specific organismal traits, which ultimately leads to the loss of trait-associated, and hence dispensable, genes by neutral decay (Albalat and Cañestro 2016; Sharma et al. 2018). Real-life examples include transitions to parasitic or symbiotic life history strategies (Spanu et al. 2010). Similarly, the long-term colonization of resource- and stimuli-deprived habitats like caves can render body coloration and vision genes dispensable (Jeffery 2009; Niemiller et al. 2013). Even entire animal clades with regressed body plans, such as nematodes and sea squirts, are characterized by an exceptional loss of ancestral gene repertoires (Spagnuolo et al. 2003; Aboobaker and Blaxter 2010; Erives 2015). Yet, there is also evidence for the loss of ancient genes without detectable correlated trait reduction and at the price of subtle fitness costs (Kawanishi et al. 2019). Even more perplexing are clade-specific extinctions of ancient genes that are otherwise indispensable for short-term survival (Breugelmans et al. 2015; Ureña et al. 2015).

As the loss of ancient genes has emerged as a consistent dimension of macroevolutionary diversity, the question arises how ancient genes transition from essentiality to dispensability. Here, we explore this issue focusing on the ancient E3 ubiquitin ligase Murinedouble minute (Mdm), which is not conserved in the model insect Drosophila melanogaster despite the conservation of its equally ancient main regulatory target, the cellular stress response transcription factor (TF) p53 in the same species (Fig. 1) (Akdemir et al. 2007; Lane and Verma 2012).

Fig. 1
figure 1

Domain conservation in, and domain interactions, between p53, Mdm, and Corp. Domain definitions based on Chakraborty et al. (2015); Ferraiuolo et al. (2016), and Åberg et al. (2017). Mdm protein domain abbreviations: p53/p63/p73BD, p53/p63/p73-binding domain; Acidic, acidic (or central) domain; zinc BD , zinc finger-binding domain; RING, really interesting new gene domain. p53 TF domain abbreviations: TAD, transactivation domain; DNA BD, DNA-binding domain; OD, oligomerization domain; SAM, sterile alpha motif; CTD, C-terminal domain; TID, transactivation inhibitory domain. Corp domain abbreviations: CMM, Corp-Mdm2 motif. Thick domain box outlines indicate protein domains that are conserved between insects and vertebrates based on primary sequence conservation. Thin domain box outlines indicate domains that are present in vertebrates but have been lost or remodeled in insects. Gray arrowed lines indicate direct domain interactions between Mdm2 and vertebrate P53 TFs that are presumed to be absent from insect p53 and Mdm homologs. Dark arrowed line indicates the direct domain interaction between the Mdm acidic domain and the p53 TF DNA BD that is possibly conserved between insects like Tribolium and vertebrates. The orange interaction lines indicate the novel direct binding of the Corp CMM-1 domain to an as of yet undefined region in the regressed p53 homolog of insect Mdm. The functionally corresponding but independently evolved P53-binding domains in mammalian Mdm2 and dipteran Corp are highlighted in orange. Protein-protein binding interactions based on Ma et al. (2006); Yu et al. (2006); Kulikov et al. (2006), and Poyurovsky et al. (2010)

Full size image

p53 transcription factors: ancient guardians of the metazoan germline

Pervasively conserved in multicellular animals (Belyi and Levine 2009; Rutkowski et al. 2010; Vilgelm et al. 2011), the p53 TF family reaches back over a billion years ago to the ancestor of Holozoa, i.e., animals and unicellular relatives (Nedelcu and Tan 2007; King et al. 2008; Rutkowski et al. 2010; Parfrey et al. 2011; Bartas et al. 2020). Most metazoans possess a singleton p53 TF homolog, which is also the case for Drosophila. Vertebrates, however, are equipped with three p53 TF family members: p53, p63, and p73 (Fig. 1) (Momand et al. 2011; Biscotti et al. 2019). Of these, paralog p53 has gained the broadest recognition due to its predominant association with tumors compared to p63 and p73. However, all three paralogs are now known to coordinate key control points of cellular health by initiating the transcription of target genes involved in DNA repair, programmed cell death, or cell cycle arrest to the effect of either rescuing or eliminating DNA-damaged cells (Joerger and Fersht 2016).

Research spanning the past 20 years has amounted to a compelling picture of functional p53 gene family evolution (Coutandin et al. 2010). Accordingly, comparative data from invertebrates compellingly suggest that the surveillance of chromosomal integrity and the elimination of handicapped cells via activation of cell death-initiating genes represent the most widely conserved, and hence likely ancestral functions of the p53 TF family (Brodsky et al. 2000; Ollmann et al. 2000; Jin et al. 2000; Schumacher et al. 2005; Lu and Abrams 2006; Derry et al. 2007; Lu et al. 2009; Rutkowski et al. 2010; Vilgelm et al. 2011), which might have originated in the context of germline cell surveillance (Rutkowski et al. 2010; Momand et al. 2011). But roles of p53 TF homologs in tissue growth regulation have also been reported for species outside vertebrates, including planarians, mollusks, and Drosophila (Pearson and Sánchez Alvarado 2010; Muttray et al. 2010; Ingaramo et al. 2018). So, future studies might reveal an even broader panel of ancestral p53 TF functions in the Metazoa.

At the molecular level, four protein regions are now unequivocally considered ancestral P53 TF family protein domains (dos Santos et al. 2016; Åberg et al. 2017). This includes the relatively large central DNA-binding domain (DNA BD) and the C-terminally adjacent oligomerization (or tetramerization) domain (OD or TD) (Fig. 1), which provides contact residues for P53 TFs to aggregate into the tetramer configuration required for stable DNA target sequence binding. Most p53 orthologs also contain a sterile alpha motif (SAM) domain, which is absent, however, from the vertebrate p53 paralog as well as from insect p53 TFs (Fig. 1) (Thanos and Bowie 1999; Lu and Abrams 2006; Åberg et al. 2017). The fourth broadly present protein domain is the N-terminal transactivation domain (TAD). It is less conserved at the primary sequence level because its multi-interactive properties are based on relatively rapidly evolving intrinsically disordered regions (dos Santos et al. 2016). The same is true for the C-terminal ends of P53 TFs, which are hubs for post-translational regulatory input (Derry et al. 2007; dos Santos et al. 2016).

These complexities notwithstanding, insect p53 TF homologs have been concluded to lack a canonical TAD in addition to the SAM domain due to the remodeling of the former and the regression of the latter during early arthropod evolution (Fig. 1) (Åberg et al. 2017). These ancient changes have implications regarding the possible conservation of P53 control mechanisms in Drosophila.

E3 ubiquitin ligase Mdm: an ancient policer of p53 transcription factor activity

Whether germline or somatic tissue, the proper execution of cellular health surveillance functions by p53 TFs hinges on their rapid dosage increase in response to cellular malfunctioning. For this reason, p53 TFs are constitutively expressed but at a low level and with a short half-life of the protein product (Lohrum and Vousden 1999). In healthy cells, a rich fabric of transcriptional, post-transcriptional, and post-translational mechanisms is in place to keep the stand-by levels of p53 below execution threshold (Love and Grossman 2012). Among the long list of factors regulating vertebrate p53, the E3 ubiquitin ligase Mdm2 stands out as a core regulator by no less than five mechanisms documented so far (Fakharzadeh et al. 1991; Momand et al. 1992; Kubbutat et al. 1997; Hu et al. 2012; Joerger and Fersht 2016): (I) operating as canonical E3 ubiquitin ligase, Mdm2 marks P53 with poly-ubiquitylation for destruction in the 26S proteasome (Hochstrasser 1996; Haupt et al. 1997; Wade et al. 2013); (II) Mono-ubiquitylation by Mdm2 modulates P53’s nuclear localization, and thus transcriptional activation potential (Li et al. 2003). (III) Mdm2 can directly disable P53 from activating target gene transcription by binding to amino acid motif FxxΦWxxL in the p53 TAD, thereby compromising its transactivation activity (Lin et al. 1994; Arva et al. 2005; Yu et al. 2006; Shin et al. 2015). (IV) Mdm2 can exert epigenetic silencing of p53 target enhancers when co-localized with P53 through the TAD (Minsky and Oren 2004). (V) Finally, Mdm2 can induce the translation of a truncated P53 isoform that neutralizes full length P53 (Yin et al. 2002; Courtois et al. 2002).

Adding another level of complexity, Mdm2 has a sister paralog in vertebrates: MdmX (also referred to as Mdm4) (Wahl et al. 2007; Wang and Jiang 2012; Wade et al. 2013). However, since Mdm4 evolved into a cofactor of Mdm2, losing endogenous ubiquitin ligase activity in the process (Huang et al. 2011), it is sufficient to reduce the comparative discussion to the vertebrate Mdm2 paralog.

The regulatory relationship between Mdm2 and p53 is deeply conserved, with evidence in mollusks and Trichoplax adhaerens, an ancient lineage of pre-bilaterian Metazoa, for a role for both factors in cell proliferation and cell survival (Lane et al. 2010a, b; Momand et al. 2011; Muttray et al. 2010; von der Chevallerie et al. 2014; Siau et al. 2016). In addition, while numerous E3 ubiquitin ligases engage in the direct regulation of specific vertebrate p53 TF homologs, i.e., p53, p63, or p73 (Jain and Barton 2010; Love and Grossman 2012; Bang et al. 2020), Mdm2 is one of few that plays roles in the regulation of all three paralogs (Zeng et al. 1999; Bálint et al. 1999; Kadakia et al. 2001; Zdzalik et al. 2010).

The regulation of p73 by Mdm2 has been studied in greater detail (Zeng et al. 1999; Bálint et al. 1999). These efforts revealed that the binding of Mdm2 to the TAD of P73 abolishes P73’s transcriptional activation capacity similar to the Mdm2–P53 interaction except for different mechanisms at the enhancer interaction level (Bálint et al. 1999). Secondly, although there is evidence for a requirement of Mdm2 in tagging P73 for proteasomal degradation (Lee and La Thangue 1999), it is apparently executed by an E3 ubiquitin ligase different from Mdm2, a difference that has been linked to the lack of a P53-like C-terminal interaction domain in P73 (Fig. 1) (Maisse et al. 2003). However, P73 has recently been found to be also contacted by Mdm2 through interaction with the SAM domain (Neira et al. 2019), which is missing in vertebrate P53 but widely conserved outside vertebrates (Åberg et al. 2017).

Another important similarity between the p53 and p73 homologs is that Mdm2 is a target of transcriptional activation by both (Fig. 1) (Lee and La Thangue 1999). The resulting gene regulatory interaction has been well studied in the case of p53 and was recognized to establish an autoregulatory feedback loop, in which p53 maintains the appropriate expression level of its own protein function repressor (Wu et al. 1993). Although P73 activates Mdm2 transcription at an approximately 2-fold lower level compared to P53 (Lee and La Thangue 1999), the same autoregulatory feedback loop seems to be in place and is thus likely ancestral.

The Drosophila gene Corp: filling in for Mdm?

Given the evidence for a deep conservation of the central role of the Mdm gene family in regulating p53 TFs and its essential nature in mouse (Jones et al. 1995; Montes de Oca Luna et al. 1995; Wahl et al. 2007), it has been surprising to find that the invertebrate models D. melanogaster and C. elegans each lack the Mdm gene family but possess functional p53 TF family homologs (Fig. 1 and Table 1) (Folberg-Blum et al. 2002; Quevedo et al. 2007; Lane et al. 2010a; Lane and Verma 2012). While the taxonomic depth of the absence of Mdm in nematodes has not been explored yet, the situation has become even more perplexing in the case of Drosophila, through the discovery that Mdm is otherwise widely conserved in the arthropods, including insects (Åberg et al. 2017). Moreover, although transgenic expression of mouse Mdm2 in Drosophila provoked abnormal wing and eye development, this occurred in the absence of detectable binding to fly homologs of vertebrate Mdm2-binding partners (Folberg-Blum et al. 2002), further escalating the question of how proper P53 TF activity levels are maintained in Drosophila.

Table 1 Conserved p53 regulators in insects and their orthologs in other animal phyla. Asterisk highlights the insect Mdm gene family, which is missing in Drosophila due to gene loss. Based on Yamasaki et al. (2007); Allton et al. (2009); Chen et al. (2011); Jain et al. (2014), and Simoes da Silva et al. (2017)
Full size table

The first proposed answer to the conundrum was based on two findings (Brodsky et al. 2004): (I) P53 levels were similar in normal and irradiated cells. (II) Phosphorylation was key to converting the Drosophila P53 TF from inactive to active form. In combination, these findings were interpreted to indicate an evolutionary shift from a combinatorial mechanism that integrates protein degradation with protein activation to a protein activation-centered mechanism (Brodsky et al. 2004). However, subsequent studies identified three E3 ubiquitin ligases that function as peripheral p53 regulators in vertebrates and are functionally conserved in Drosophila (Table 1). This includes the E3 ubiquitin ligase Synoviolin (Yamasaki et al. 2007), Sexcombs extra (Sce), the Drosophila homolog of vertebrate E3-ubiquitin ligase ring finger protein 2 (Ring1b/RNF2) (Simoes da Silva et al. 2017) and Drosophilabonus (bon), which is a homolog of vertebrate Tripartite-motif protein 24 (Trim24) (Beckstead et al. 2001; Allton et al. 2009; Ying et al. 2011). Further intriguing is the fact that vertebrate Trim24 is transcriptionally activated by p53, thus engaging in the same type of autoregulatory feedback loop as vertebrate Mdm2 with p53 (Jain et al. 2014). While it is not yet known whether this is also the case in Drosophila, bon was proposed to represent a candidate replacement for Mdm in Drosophila (Allton et al. 2009). These ambiguities aside, it is now clear that Drosophila P53 levels are similarly regulated by E3 ubiquitin ligase-based mechanisms as in vertebrates in addition to its activation by phosphorylation (Brodsky et al. 2004).

While it is reasonable to speculate that any of the above E3 ubiquitin ligases might have filled in for the lack of Mdm in Drosophila, it is also important to keep in mind that the coexistence of all of these regulators for hundreds of millions of years together with Mdm suggests a high selection pressure on maintaining them in parallel. Nonetheless, Mdm might have assumed less critical significance in insects before its loss during dipteran evolution, for instance, through incremental functional compensation by p53-regulating E3 ubiquitin ligases such as bon, Sce, or Synoviolin. To probe for this possibility, we conducted gene expression and knockdown experiments in the red flour beetle Tribolium castaneum, where all of these gene families are conserved in addition to Mdm (Table 1).

In a second line of investigation, we focused on a fifth p53 regulator of Drosophila that has been noted to display mechanistic similarities to vertebrate Mdm2. Named Companionof reaper (Corp), based on being transcriptionally activated by p53 like the pro-apoptotic Drosophila gene reaper (rpr) (Brodsky et al. 2000), Corp has been proposed to represent a “functional analog” of vertebrate Mdm2 (Chakraborty et al. 2015). The Corp gene was discovered by virtue of its capacity to decrease P53 protein levels in DNA-damaged cells (Chakraborty et al. 2015). While the absence of E3 ubiquitin ligase signature protein domains in Corp implied that Corp is not a member of the Mdm gene family (Fig. 1), Corp was found to possess a domain with direct P53-binding capacity, which shares protein sequence similarities with the p53/p63/p73-binding domain (p53/p63/p73BD) of vertebrate Mdm2 (Chakraborty et al. 2015). Moreover, Corp was discovered to be transcriptionally activated by p53, thus engaging in the same type of feedback loop that maintains the steady-state levels of Mdm2 and p53 in vertebrates (Chakraborty et al. 2015). The possibility that Corp represents a functional analog of Mdm2 has been considered by others, but it was cautioned that further studies were needed to understand how the presumed loss of Mdm relates to the evolution of Corp (Ingaramo et al. 2018). In the absence of information on Corp conservation beyond Drosophila species, we explored whether the loss of Mdm in Drosophila might be related to the emergence of Corp as compensatory gene activity by surveying the evolutionary conservation of both gene families in the dipteran genome sequence and transcriptome databases.

Materials and methods

Animal culture

The Tribolium castaneum Georgia-I (GA-I) strain was used for cloning, RT-PCR and knockdown experiments. All life cycle stages were maintained in constant darkness at 31 °C in whole wheat flour enriched with 5% dry baker’s yeast and 0.5% fumagillin. Experimental animals were cultured in unbleached white flour likewise enriched with 5% dry baker’s yeast and 0.5% fumagillin.

Homolog compilation and protein sequence conservation analysis

Whole genome shotgun contigs (wgs) and transcriptome shotgun assemblies (TSA) were searched using the tBLASTn module in the basic local alignment search tool (BLAST) suite of the National Center of Biotechnology Information (NCBI) using the protein sequences of mouse Mdm2 (U47934) and Drosophila Corp (CG10965) as queries (Altschul et al. 1990). Candidate orthologs were scrutinized by the reciprocal best BLAST hit approach (Hirsh and Fraser 2001; Jordan et al. 2002; Wall et al. 2003). Multiple protein sequence alignments were generated with T-Coffee (Notredame et al. 2000). All accession numbers and the protein sequences of Mdm and Corp compiled for this study are available in Supplementary Data files 2, 3, 4.

Semi-quantitative RT-PCR

Total RNA was extracted from embryos, larval head tissue, larval trunk tissue, whole pupae, adult head tissue, and adult abdomens using the RNAqueous-R midi kit (Ambion). Template cDNA was generated by reverse transcription with the Retroscript-R kit (Ambion). PCR amplification was carried out with an Eppendorf Mastercycler ep Gradient 5341. For cDNA input normalization, we amplified the housekeeping gene armadillo 1 from Tribolium (LOC659291). The sequences of all primers used are listed in Supplementary Data file 1. Cycling conditions: 2 min initial denaturation followed by 23 cycles of 30 s denaturation at 94 °C, 40 s annealing at 50 °C, and 60 s elongation at 72 °C.

RNA interference

Three different dsRNAs matching non-overlapping regions in the coding region of the Tribolium Mdm homolog (LOC664204) were generated by in vitro transcription from PCR-generated template DNAs: Tcas Mdm147–525, Tcas Mdm227–503, and Tcas Mdm934–1084 (numbers in superscript correspond to nucleotide position in the coding sequence). The sequences of the primers used for template DNA production including those generating the EGFP dsRNA fragment for the negative control knockdown injections are provided in Supplementary Data file 1. All dsRNAs were synthesized by bidirectional in vitro transcription using the MegaScript T7 transcription kit (Ambion). dsRNA yields were examined by gel electrophoresis and concentrations measured using a NanoDrop 1000 spectrophotometer. dsRNAs were diluted in water to a concentration of 1 μg/ul and injected together with phenol red into the abdomens of adult virgin females, which were anesthetized by CO2 and positioned laterally on a slide using double-stick tape.

Analysis of egg production and embryonic viability

Female pupae were collected to generate 10 animal large experimental populations of virgin females, which were injected with buffer, EGFP dsRNA or Mdm dsRNA and cultured on unbleached white flour. Animals that died within the first 24 h after injection were excluded from viability analyses and two males were added to each culture. Thereafter, eggs were collected every 24 h to monitor daily per female egg production rate and cultured at 31 °C in separate cohorts to monitor the proportion of larval hatching.

Results

Mdm is essential for normal adult survival, egg production, and embryonic viability in Tribolium

In vertebrates, loss or reduction of Mdm2 results in embryonic lethality due to the increase in cellular P53 (Jones et al. 1995; Montes de Oca Luna et al. 1995). The lack of the Mdm gene family in Drosophila, despite the conservation of p53, therefore raised the question of how the loss of such an important p53 regulator had been tolerated during fly evolution. One possible explanation was a decoupling of the p53–Mdm regulatory axis during an earlier phase of insect evolution, rendering Mdm non-essential.

To probe whether insect Mdm homologs still provide functions that are essential for organismal survival, we conducted experiments in the red flour beetle T. castaneum, an experimentally versatile model species (Bucher et al. 2002) with a well-curated genome sequence draft (Richards et al. 2008; Herndon et al. 2020). Previous searches identified Mdm homologs in many arthropod species (Åberg et al. 2017). Leveraging this information, we cloned regions of the singleton Mdm homolog of Tribolium, which is characterized by a well-conserved zinc finger-binding domain (zinc BD) and RING domain, which are diagnostic for E3 ubiquitin ligases, in addition to the acidic domain (Figs. 1 and 2a). Semi-quantitative RT-PCR analysis indicated equivalent expression levels of Tribolium Mdm at all major life cycle stages (embryo, larva, pupa, and adult) and in different tissue regions (head and trunk) (Fig. 2b), consistent with the characteristics of a housekeeping gene. Adult animals injected with Mdm dsRNA invariably died within 20 days in contrast to control animals injected either with buffer only or EGFP dsRNA. This result was reproduced in three replicates with non-overlapping dsRNAs to rule out off-target artifacts (Fig. 2c). The average life span post-injection was 8.6 (± 1.0) days. Tracking egg-laying over the course of 15 days revealed an over 50% lower daily egg production rate in an Mdm dsRNA injected cohort of 10 females in comparison to equally large control populations of females injected with EGFP dsRNA or buffer (Fig. 2d). Tracking larval hatching in the same three populations as a proxy of embryonic viability revealed an approximately 50% lower proportion of eggs that produced larvae in the Mdm knockdown population compared to the EGFP dsRNA or buffer-injected populations (Fig. 2e).

Fig. 2
figure 2

Functional analysis of Tribolium Mdm. a Sequence conservation in the ring finger domain and zinc finger domain of Tribolium Mdm. Black and gray sidebars indicate insect and other arthropod species, respectively. Diagnostic cysteine residues indicated by gray background shading. Species abbreviations: Aaeg, Aedes aegypti; Bmori, Bombyx mori; Dmel, Drosophila melanogaster; Isca, Ixodes_scapularis; Lpol, Limulus polyphemus; Mmus, Mus musculus; Nvit, Nasonia vitripennis; Rpro, Rhodnius prolixus; Smim, Stegodyphus mimosarum; Tadh, Trichoplax adhaerens; Tcas, Tribolium castaneum; Dpul, Daphnia pulex. b Semiquantitative RT-PCR expression analysis. Top gel: Mdm RT-PCR bands resulting from amplification with primer combination Tcas_Mdm_A3 and Tcas_Mdm_B3. Bottom lane: Armadillo 1 RT-PCR bands for measuring cDNA input amplified with primer combination Tcas_armB1 and Tcas_armB4. Sample codes: M, invitrogen 1 kb ladder DNA marker; E, embryonic; LH, larval head; LT, larval trunk; P, whole pupa; AH, adult head; AA, adult abdomen; −, negative control. c Survival curves of three populations of 20 adult female Tribolium each injected with one of three non-overlapping Mdm dsRNA fragments at 1 μg/ul concentration: Tcas Mdm147–525 (n = 20, Tcas Mdm227–503 (n = 20) and Tcas Mdm934–1084 (n = 19), adult females injected with EGFP dsRNA at 1 μg/ul concentration (hatched lines; n = 20) or with injection buffer only (solid line; n = 42). d Egg production in Mdm knockdown females (n = 9) compared to buffer (n = 9) or EGFP dsRNA (n = 9) control injected females. Daily per female egg production over a time span of 15 days starting from day 4 after injection was significantly different between the Mdm dsRNA injected females and either control based on two-tailed t test results (Mdm dsRNA vs. EGFP: p = 0.00001; Mdm dsRNA vs. buffer: p = 0.005; EGFP vs. buffer: p = 0.7). Error bars represent standard deviations. e Embryonic viability over a time span of 13 days starting from day 5 after injection measured as daily larval hatching proportion in eggs from Mdm knockdown females compared to EGFP control injected females (Mdm dsRNA vs. EGFP: p = 0.01; Mdm dsRNA vs. buffer: p = 0.01; EGFP vs. buffer: p = 0.85). Error bars represent standard deviations

Full size image

Taken together, these findings characterized Mdm as an essential gene in Tribolium that was also required for normal female germline activity and embryonic development. Although representing only a single, distantly related data point, this result spoke against the idea that Mdm assumed reduced functional significance prior to its extinction in the Diptera.

Complementing but partially overlapping phylogenetic distributions of Mdm and Corp in brachyceran Diptera

Given the evidence from Tribolium that Mdm has remained essential in insect species where it has remained conserved, the possibility that the p53 regulator Corp might have rendered Mdm dispensable in dipteran insects continued to be an attractive idea. From an evolutionary perspective, this scenario would be supported by the lack of Corp in dipteran species in which Mdm has remained conserved and the complementary presence of Corp in dipteran species that lack Mdm.

To elucidate the distributions of Mdm and Corp in the dipteran tree of life, we conducted homolog searches in dipteran genome and transcriptome databases (Fig. 3 and Supplementary data file 2) (Scott et al. 2014; Anstead et al. 2016; Papanicolaou et al. 2016). This effort identified Mdm homologs in 15 dipteran species from 10 distantly related families representing many ancient suborders (Culicidae, Keroplatidae, Dolichopodidae, Platypezidae, Phoridae, Tabanidae, Pediciidae, Psychodidae, Asilidae, Trichoceridae). Based on this taxonomic distribution, we were able to conclude that the Mdm gene family has remained conserved in many dipteran lineages but went extinct approximately 95–85 million years ago in the lineage to the last common ancestor of Pipunculidae (big-headed flies) and their large sister clade, the Schizophora, which includes Drosophila (Fig. 3) (Pauli et al. 2018; Wiegmann et al. 2011).

Fig. 3
figure 3

Conservation of Mdm and Corp homologs in the Diptera. Numbers in parentheses following family names reflect the number of species in which Mdm or Corp homologs have been detected. Green background highlights the phylogenetic range of families with species in which only Mdm homologs were detectable. Blue background highlights the phylogenetic range of families with species in which only Corp homologs were detectable. Light brown background highlights phylogenetic range of families in which species with both Mdm and Corp homologs were detected. For homolog and species details see supplementary data. Phylogenetic relationships and time scale based on Wiegmann et al. (2011); Caravas and Friedrich (2013), and Pauli et al. (2018)

Full size image

Consistent with a complementary taxonomic distribution of Mdm and Corp, we found homologs of Corp in all of the seven species from four schizophoran families (Drosophilidae, Tephritidae, Muscidae, Calliphoridae) where we failed to detect Mdm. Moreover, the same was the case for the one representative species that could be sampled from the Pipunculidae (Fig. 3). However, we also detected Corp homologs in species where we had found Mdm. This included four species of hoverflies (Syrphidae), with one of them, Leucozona lucorum, possessing two homologs of Corp (Supplementary data files 2, 3, 4). Moreover, we found partial but unambiguous homologs of both Mdm and Corp in one species of pointed-wing flies (Lonchopteridae) (Fig. 3 and Supplementary data files 2, 3, 4). The latter finding was significant by dating the origin of the Corp gene back to the early diversification of cyclorrhaphan Diptera, approximately 150 million years ago (Fig. 3).

Most importantly, these findings revealed that the emergence of Corp did not result in an immediate loss of Mdm. Instead, the conservation of both genes in present-day Lonchopteridae implies their coexistence for about 150 million years in this ancient lineage of cyclorrhaphan Diptera. Consistent with this unambiguous evidence for the mutual compatibility of the two p53 regulators, their phylogenetic distribution also implies that the origin of Corp was followed by its coexistence with Mdm for at least 50 million years before the loss in the ancestor of the Pipunculidae and the Schizophora (Fig. 3).

Genomic evidence of Corp de novo birth in the CG1632 locus

The lack of detectable Corp homologs outside cyclorrhaphan Diptera suggested that Corp originated via the transcriptional activation of previously non-transcribed sequence region, i.e., de novo, during the early evolution of cyclorrhaphan flies, approximately 150 million years ago (Fig. 3). This was further supported by the absence of known protein domains in the Corp gene product based on searches in the Pfam database (Finn et al. 2014), which argued against gene duplication or exon shuffling as origination mechanisms. In the case of evolutionarily young de novo originated genes, the absence of a homolog in the corresponding genome regions of closely related outgroup species adds further support for the de novo origination of a new gene (Schlötterer 2015; McLysaght and Hurst 2016). This criterion is usually not applicable to older de novo originated genes due to the loss of homologous sequence similarity in neutrally evolving sequence regions between distantly related species. However, genetic linkage can be stable over long evolutionary time spans. Conserved linkage has therefore the potential to define the genomic place of origin of a de novo gene that can be compared to outgroup species that are presumed to lack the gene.

In Drosophila, Corp is situated in the last intron of the still uncharacterized locus CG1632, a homolog of the Corin serine protease gene family based on preliminary BLAST results (Yan et al. 1999). To explore whether this linkage constellation is conserved, we examined the genomic environments of Corp homologs in the four dipteran species with available genome assemblies: Drosophila virilis, Lucilia cuprina (Calliphoridae), Musca domestica (Muscidae), and Ceratitis capitata (Tephritidae). This effort revealed that Corp is present in the homologous intron number 6 of the CG1632 locus in all of these species (Fig. 4 and Supplementary data file 5). Moreover, the opposite coding orientation of Corp in relation to the CG1632 locus was also conserved. This was even true for a pair of tandem duplicated Corp sister paralogs in the Mediterranean fruit fly C. capitata that we discovered in addition to the independent Corp duplication in the hoverfly species L. lucorum (Fig. 4 and Supplementary data file 2).

Fig. 4
figure 4

Conserved linkage of Corp and Corin paralog CG1632. Schematic comparison of the exon organization of the Drosophila locus CG1632, revealing the conserved linkage of Corp in the last intron of CG1632. Exon, intron, and intergenic regions are not drawn to scale. Homologous genes are highlighted by identical box colors. Locus LOC116344225 of the gall midge species C. nasturtii (Cecidomyiidae) lacks a homolog in D. melanogaster. See Supplementary data file 5 for species details and NCBI genome browser links

Full size image

The last introns of the CG1632 homologs in the gall midge species Contarinia nasturtii (Cecidomyiidae), the mosquito species Anopheles gambiae and Culex pipiens (Culicidae) (Fig. 4), and Tribolium (not shown), however, lack evidence of additionally transcribed sequence regions based on available RNAseq profiles (Supplementary data file 5). Although limited to schizophoran Diptera, these findings bolstered the case for a de novo origin of Corp by pinpointing the last intron of CG1632 as its bona fide genomic place of birth.

Protein sequence conservation in Corp

Vertebrate Mdm proteins are about 500 amino acids in length and characterized by four domains: The p53/p63/p73BD, the acidic domain, the zinc BD, and the RING domain (Fig. 1) (Hu et al. 2012). Insect Mdm homologs are shorter, ranging between 370 and 430 amino acids in length, reflecting the loss of the N-terminal p53/p63/p73BD (Fig. 1, Supplementary data file 5) (Åberg et al. 2017). The sizes of the conceptual Corp gene products that we were able to compile from a larger range of dipteran species, by contrast, were substantially shorter, ranging only between 160 and 200 amino acids (Supplementary data file 4). This discrepancy raised the question of how the structural and functional complexity of Mdm was matched by the seemingly less complex Corp protein homologs.

Using bioinformatic strategies, Chakraborty et al. (2015) identified two putative p53-binding domains in Corp based on sequence similarity to protein sequence motifs in vertebrate Mdm2. Only one of them, however, termed Corp-Mdm2 motif 1 (CMM-1), was found to effectively bind to Drosophila P53 (Chakraborty et al. 2015). Multiple sequence alignments of the extended sample of Corp protein sequences from dipteran species revealed that both previously defined Corp domains, CMM-1 and CMM-2, are highly conserved and embedded in an 80 amino acids long block of overall high sequence conservation (Fig. 5).

Fig. 5
figure 5

P53-binding domain conservation in Corp. Multiple protein sequence alignment of dipteran Corp homologs. Highlighted in bold print: the CMM-1 and CMM-2 domains previously defined in Drosophila Corp (Chakraborty et al. 2015). Sequence conservation color code based on T-Coffee alignment algorithm (Wallace et al. 2006; Taly et al. 2011): Red = high, yellow = intermediate, green or purple = low. Species abbreviations: Belon, Baccha elongata (Syrphidae); Dmel, Drosophila melanogaster (Drosophilidae); Eper, Eristalis pertinax (Syrphidae); Lcup, Lucilia cuprina (Calliphoridae); Lluc, Leucozona lucorum (Syrphidae); Mdom, Musca domestica (Muscidae); Msca, Melanostoma scalare (Syrphidae); Pnoc, Pipiza noctiluca (Syrphidae); Rzep, Rhagoletis zephyria (Tephritidae); Scal, Stomoxys calcitrans (Muscidae)

Full size image

In addition, we identified a 17 amino acids long region of high sequence conservation here termed Corp motif 1 (CM-1) that is likely to exert a conserved function (Fig. 5). Moreover, we noted that the C-terminal amino acids SIRI appear equally highly conserved. These findings define new regions of interest for future functional study, for instance, regarding the evidence from large scale protein-protein interaction surveys that Corp does engage with additional proteins besides the Drosophila P53 TF (Callaghan et al. 1998; Guruharsha et al. 2011; Chakraborty et al. 2015).

Discussion

Possible functions of Mdm in Tribolium

Our gene expression and gene knockdown analyses in Tribolium revealed that Mdm is constitutively expressed throughout all life cycle stages and required for normal adult survival, egg production, and embryonic development. These findings speak against the possibility that the loss of Mdm in dipteran insects may have been due, at least partly, to a less critical role of Mdm for organismal functioning in insects. It is less clear yet, however, whether the effects we observed in Mdm knockdown Tribolium are compatible with the expected downstream effect of P53 TF upregulation. Although insect Mdm homologs including that of Tribolium are characterized by the lack of the p53/p63/p73BD, the acidic domain remains a candidate domain for a direct interaction with P53 TF homologs (Fig. 1). This conjecture is based on the evidence of direct binding between the acidic domain of Mdm2 to the DNA BD of p53 in vertebrates (Figs. 1 and 5) (Ma et al. 2006; Yu et al. 2006; Kulikov et al. 2006).

The decrease of egg production in Mdm knockdown Tribolium is reminiscent of the effect of misregulated p53 TF levels in germline guarding. Also, the evidence of compromised embryonic development in Mdm knockdown Tribolium may parallel the lethality of Mdm2-deficient mouse embryos (Jones et al. 1995; Montes de Oca Luna et al. 1995). However, increased p53 levels had a life span prolonging effect in female Drosophila (Waskar et al. 2009), which does not align well with the shortened life span of the Mdm knockdown Tribolium females in our experiments. Another phenotype of strong p53 overexpression in Drosophila is the triggering of dramatic apoptosis in developing tissue (Ollmann et al. 2000; Jin et al. 2000). Again, this effect seems unlikely to be at play in the Tribolium Mdm knockdowns as rampant apoptosis might be expected to lead to rapid tissue decline, which does not seem to be compatible with both the slow progression of adult lethality as well as the intermediate penetrance of the embryonic and egg production phenotypes in Mdm knockdown Tribolium.

The plethora of newly discovered “non-canonical functions” of Drosophila p53 in the regulation of tissue development (Ingaramo et al. 2018) further complicates the issue of developing an expected phenotype profile related to global Mdm and p53 TF misregulation in a distantly related insect like Tribolium. Quantitative analyses of apoptosis and p53 expression levels will therefore be needed to probe for regulatory interactions between Mdm levels and P53 TF activity in Tribolium.

In mice, the embryonic lethality of Mdm2-deficient offspring is primarily due to the upregulation of p53 (Jones et al. 1995; Montes de Oca Luna et al. 1995) Mdm2-deficient mouse embryos fail to complete development while Mdm2 + p53 double deficient mice produce viable offspring. In principle, the same genetic strategy could be applied to test for the p53-dependence of the Mdm knockdown phenotype in Tribolium. We abstained from expanding the scope of our study in this direction, as Tribolium possesses a tandem duplicated pair of p53 TF homologs (LOC657307, LOC655327) (Åberg et al. 2017), which are expressed at similar levels based on available RNAseq evidence (Herndon et al. 2020). As these paralogs may be fully or partially redundant, probing the epistatic relationship between p53 TF activity and Mdm will likely require a more expansive set of combinatorial gene knockdown or gene editing experiments.

Moreover, the presence of other possibly p53 regulating E3 ubiquitin ligases in Tribolium, i.e., bon, Sce, and sip3 (Table 1), increases the number of possible regulatory interactions to explore. Given the interaction of Mdm2 with other pathways in vertebrates (Bouska and Eischen 2009), it is also conceivable that the effects of Mdm knockdown in Tribolium are the result of misregulated mechanisms that are independent of p53. And finally, there may be Mdm-associated mechanisms in insects or beetles that have no precedent in other organismal groups. We therefore suspect that it will require a broad range of experiments to unravel the molecular cause or causes underlying the essentiality of Mdm in Tribolium.

Corp: a de novo originated generic p53 antagonist in cyclorrhaphan Diptera

The Corp gene was identified in a targeted screen for regulators of Drosophila p53 TF activity in its role as a cell death inducer in response to DNA damage (Chakraborty et al. 2015). The combined evidence from genetic, bioinformatic, and biochemical studies led to the conclusion that Corp represents a “functional analog” of vertebrate Mdm2 (Chakraborty et al. 2015). This terminology reflects four aspects of how Drosophila Corp and vertebrate Mdm2 compare. Three of them are similarities: (I) Both factors reduce P53 protein levels. (II) The expression of both gene products is regulated by an autoregulatory feedback loop that involves p53 as their direct transcriptional activator of its own protein function repressor. (III) The P53 TF-binding regions of the Drosophila Corp and Mdm2 proteins are characterized by significant sequence similarities. Contrasting these similarities, Corp differs from Mdm2 by the absence of all E3 ubiquitin ligase signature domains, ultimately identifying the two genes as members of phylogenetically unrelated gene families. It is the last aspect, i.e., the memberships to different gene families, which constitutes the prime evidence that the similarities of Corp and Mdm2 are not explained by the shared descent but convergent acquisitions, thus defining them as “functional analogs” (Chakraborty et al. 2015).

Of note, there is a lesser explored but equally informative discrepancy between Corp and Mdm that speaks to their belonging to different gene families. Drosophila Corp and vertebrate Mdm2 have been found to possess two regions of significant sequence similarity (Chakraborty et al. 2015). Of these, that between CMM-1 of Corp and a region in p53/p63/p73BD of vertebrate Mdm2 has been found to be of functional significance as both of them have the capacity to bind their species-specific P53 homologs (Fig. 1) (Chakraborty et al. 2015). The p53/p63/p73BD of vertebrate Mdm2, however, contacts the canonical TAD of P53 (Lin et al. 1994; Arva et al. 2005; Yu et al. 2006; Shin et al. 2015), which, importantly, seems to be lacking in the p53 TF homologs of insects, including Drosophila (Fig. 1) (Poyurovsky et al. 2010; Åberg et al. 2017). This leads to the conclusion that the sequence similarities between the CMM-1 of Drosophila Corp and the p53/p63/p73BD domain of vertebrate Mdm2 are coincidental since Corp CMM-1 must be interacting with a target region in Drosophila P53 that, while still undefined, is different from the TAD of vertebrate P53.

Further consistent with a lack of evolutionary ancestry between the Mdm and Corp gene families, our genomic studies build a strong case that Corp originated de novo in cyclorrhaphan Diptera in contrast to the metazoan ancientness of Mdm. Recognizing the independent origins of the two gene families leads to the conclusion that Drosophila Corp acquired not only one but two functional similarities convergently with vertebrate Mdm2 (Chakraborty et al. 2015): (I) a direct role in the negative control of p53 protein levels and (II) being transcriptionally activated by p53. While intriguing, it is important to note that these similarities are not Mdm2-specific. While Mdm2 is the major regulator of vertebrate p53, there is a large number of additional E3 ubiquitin ligases that contribute to the regulation of vertebrate p53 besides Mdm2 (Harris and Levine 2005). Importantly, some of them have been found to engage in the same type of autoregulatory feedback regulation with p53, by being transcriptionally activated by the latter (Leng et al. 2003; Dornan et al. 2004). This is, for instance, true for E3 ubiquitin ligase Trim24, the vertebrate homolog of Drosophila bon, which has likewise been proposed to represent the replacement for Mdm (Allton et al. 2009; Jain et al. 2014). In addition, even post-translational regulators of p53 that are different from E3 ubiquitin ligases are targets of transcriptional activation by p53 in vertebrates (Bang et al. 2020).

The multiplicity of similarly wired autoregulatory feedback loops in the control of p53 indicates strong constraints on the logic of p53 control. This may be part of the network requirements that facilitate the fast switch from steady-state to escalated P53 activity levels in response to cell stress signal (Wang et al. 2018). In recognition of this context, it seems that Corp represents less of a specifically Mdm2-like factor but more of a generic negative regulator of p53 TF activity.

Incremental regression and clade-specific loss of Mdm during arthropod evolution

Our study was motivated by the question of whether the loss of Mdm in a part of the dipteran tree of life which includes Drosophila might have been triggered by compensatory replacement through Corp. Our finding of the near complementary de novo emergence of Corp and extinction of Mdm during the diversification of cyclorrhaphan Diptera leaves room for the possibility of a compensatory role of Corp in the trajectory that culminated in the dispensability of Mdm. The combined evidence, however, suggests a scenario in which the loss of Mdm in dipteran insects was the outcome of incremental structural and regulatory regression. In support of this idea, previous studies uncovered evidence of an early partial regression of the regulatory scope of the Mdm gene in arthropods (Fig. 6) (Åberg et al. 2017). Specifically, this change was marked by the reduction of the N-terminal p53/p63/p73BD of Mdm in the last common ancestor of insects and crustaceans, about 500 million years ago (Åberg et al. 2017), correlated with the corresponding loss of its binding targets, i.e., the TAD and the SAM domains, in the p53 homologs of the same arthropod groups (Fig. 1) (Chen et al. 1993; Ollmann et al. 2000; Folberg-Blum et al. 2002; Poyurovsky et al. 2010; Åberg et al. 2017; Neira et al. 2019).

Fig. 6
figure 6

Evolutionary conservation, gain, and loss of p53 transcription factor regulating genes. Stop signs indicate gene loss events. Starburst sign indicates the de novo origin of the dipteran Corp gene family. The relative time points of the Mdm and Trim24 losses in C. elegans are arbitrarily placed. The gene tree also reflects the vertebrate duplication of the Mdm gene family, generating Mdm2 and MdmX (Wade et al. 2013)

Full size image

In dipteran insects, the insect/crustacean loss of the p53/p63/p73 BD in Mdm was followed by the de novo emergence of Corp approximately 150 million years ago. This event minimally supplied novel Mdm-like p53 regulatory activities related to DNA-damage triggered cell death regulation in germline cells. A 1:1 compensatory gene loss and gain scenario, however, seems unlikely for a number of reasons. Most importantly, our findings reveal that the emergence of Corp was insufficient to enforce or allow for the loss of Mdm in select cyclorrhaphan lineages until present time, i.e., hoverflies (Syrphidae) and pointed-wing flies (Lonchopteridae), which possess both factors. Moreover, even in the lineage to the last common ancestor of Pipunculidae and schizophoran Diptera, where Mdm ultimately faced extinction, Mdm coexisted with Corp for possibly up to 50 million years. Given the importance of a robust control of p53 levels, it is not unreasonable to suspect genetically redundant p53 TF control pathways to persist over long evolutionary time scales. Indeed, this speculation is supported by the conservation of multiple ancient p53 regulators in arthropods (Table 1 and Fig. 5), some of which are suspected to be partially redundant (Chakraborty et al. 2015).

In place of a 1:1 compensatory gene loss and gain scenario, we speculate that additional compensatory gene gain events paved the way towards the extinction of the Mdm gene family in early schizophoran Diptera. In support of this, Drosophila Corp seems specifically involved in the control of p53 activity in germline DNA-damaged cells (Chakraborty et al. 2015). This is consistent with its elevated, if not specific, expression in the female germline cell population based on modENCODE expression data (Chen et al. 2014). Vertebrate Mdm genes, in contrast, also execute cell curation functions that are independent of p53 (Bouska and Eischen 2009; Hu et al. 2012; Wade et al. 2013), which aligns well with the multiple requirements we found for Mdm in Tribolium.

In summary, while our findings are compatible with a scenario in which Corp replaced specific Mdm functions in the context of p53 regulation, it remains to be seen whether additional Mdm functions existed that were replaced by additional compensatory changes. These hypothesized additional ancestral p53-related and independent functions could be probed in experimental insect model species that preserved Mdm in the absence of Corp such as Tribolium. Likewise, the question of whether and to which degree Mdm and Corp execute unique vs. redundant functions could be clarified in dipteran model species that possess both genes such as the hoverfly species Episyrphus balteatus (Rafiqi et al. 2008). These open lines of investigation notwithstanding, we propose that early terminal domain reductions followed by the de novo evolution of Corp represent milestones in the incremental regression of ancestral pleiotropy of the ancient and essential Mdm gene family, which culminated in its loss in the lineage to schizophoran Diptera (Fig. 6).

Of note, this incremental regression scenario is also consistent with the scarcity of Mdm gene family extinction events in the animal tree of life, as it fits the low odds of a multi-event contingent outcome. So far, the only additional documented case of Mdm gene family absence is C. elegans (Fig. 6) (Quevedo et al. 2007; Lane et al. 2010a; Lane and Verma 2012), which offers the opportunity to test whether a similarly incremental series of compensatory gene function loss and gain events culminated in the nematode loss of the Mdm gene family. Based on what we were able to learn from studying Mdm’s extinction in relation to the de novo origin of Corp during dipteran evolution, such studies are poised to deliver deeper insights into the roles of origination and extinction events in the evolutionary turnover of essential genes and the rewiring of ancient molecular pathways.